Feedforward compensation can reduce dynamic errors and improve tracking accuracy and response speed, but residual errors often exist in engineering applications. This study proposes a jerk-segmented adaptive iterative learning control (ILC) strategy to address this limitation. The surrogate model's feedforward compensation generates high-quality initial inputs. The trajectory is divided into segments based on jerk. Within each segment, tracking errors are filtered using a zero-phase Butterworth filter to extract their low-frequency trend. A nonlinear PID-type learning law is then applied: the proportional term acts on the raw error with nonlinear scaling and saturation, while the integral and derivative terms operate on the filtered trend to improve stability. All control parameters are adaptively adjusted based on the filtered error's energy. Simulation and experimental results under 10 kHz sampling show that the proposed method significantly improves tracking accuracy. Compared with norm-optimal ILC and the case without ILC, the technique reduces steady-state RMSE by up to 57.40 % and peak tracking error by up to 38.19 % during high-dynamic motion phases. The proposed method ensures consistent performance across simulation and physical experiments.
扫码关注我们
求助内容:
应助结果提醒方式:
